A backtest is testing two systems: the signal and the size.

Two backtests can use identical signals and produce very different drawdowns, returns, and failure modes because they size positions differently. If the sizing rule is vague, the performance report cannot tell you whether the result came from the strategy idea, hidden leverage, or concentrated exposure.

Notional exposure is not the same as risk

Notional exposure is the market value controlled by a position. Risk is the loss the strategy is prepared to accept under a defined scenario. A $10,000 position with a nearby exit may have a different planned loss than a $10,000 position with no protective exit, although both have the same notional value.

The planned loss is not a guarantee. Gaps, slippage, liquidity, order behavior, and fast markets can produce a larger outcome. Still, separating notional size from intended risk makes the rule inspectable.

Common sizing approaches

Fixed quantity

Every trade uses the same number of shares, contracts, or units. This is simple to reproduce, but the dollar exposure changes as price changes and the risk can vary when exit distances differ.

Fixed notional value

Each trade receives the same dollar allocation. This normalizes market value more than fixed quantity, but it still does not make the intended loss equal across trades.

Percentage of equity

The position uses a fixed share of account equity. Because the base changes after gains and losses, compounding becomes part of the result. Record whether the test uses starting equity, current equity, or another base.

Risk-based sizing

Quantity is derived from a planned risk amount and a defined distance between entry and the risk reference. The approach can make planned losses more comparable, but only if the reference, rounding, gaps, and maximum exposure are documented.

Volatility-based sizing

Exposure changes with a volatility estimate. Quieter conditions may receive more units and volatile conditions fewer. The indicator lookback, timing, floor, cap, and behavior during sudden volatility changes all belong in the specification.

Position size is not formatting applied after the strategy. It is one of the strategy's rules.

Write the complete sizing formula

A reproducible sizing rule states the account value used, the risk fraction or allocation, the price or volatility input, rounding, minimum size, maximum size, leverage limit, and what happens when the required quantity cannot be traded.

Also define when the size is calculated. If the formula uses a closing price or volatility value, confirm that the input existed before the order. The look-ahead timing checklist applies to sizing inputs as well as entry signals.

Five distortions to look for

1. Hidden leverage. A test may allow exposure beyond account equity without recording financing, margin, or liquidation constraints.
2. Fractional or impossible units. The calculation may produce quantities the instrument or account could not trade.
3. Unbounded concentration. One position, symbol, sector, or correlated group may control most of the outcome.
4. Perfect exits. The sizing formula may assume a planned stop always fills at the exact reference price.
5. Inconsistent compounding. Gains may increase future size while losses, withdrawals, or overlapping trades are handled differently.

Test sizing independently from the signal

Hold the entry and exit rules constant while comparing a small, defensible set of sizing policies. The goal is not to search for the largest historical return. It is to understand how the risk rule changes drawdown, concentration, turnover, leverage, and dependence on a few trades.

Remove the largest winner, review the largest loss, and inspect periods with several simultaneous positions. If a small number of oversized trades controls the conclusion, report the result as concentrated.

Portfolio risk can exceed per-trade risk

A rule that limits each trade separately may still create a large combined exposure. Five positions can respond to the same market event, especially when instruments or signals are correlated. Record the maximum number of open positions, total notional exposure, combined planned risk, and any group-level cap.

Overlapping positions also affect cash availability and the order in which opportunities can be taken. A backtest that assumes every valid signal receives full size may be allocating the same capital more than once.

Stress the assumptions around the size

Position size interacts with execution. Larger orders can face wider slippage and market impact, while exits may fail to occur at the planned price. Use the trading-cost stress-test framework to connect quantity with spread, liquidity, delay, and fill assumptions.

Then apply the broader backtest validation checklist. A transparent sizing rule does not validate the signal, but it prevents size from hiding what the signal actually did.

The useful conclusion

A position-sizing test should tell you how quantity is derived, whether exposure stays within documented caps, which trades dominate the result, and how combined risk behaves. It cannot guarantee a future loss limit or execution price.

Preserve the sizing formula and its inputs with every result. Otherwise, a later comparison may be evaluating a different risk system while pretending only the entry rule changed.